malloc(0)

[Roland McGrath]

> I was thinking that (a) it would flush out bugs, and (b) a program written
> to assume that realloc(ptr, 0) is equivalent to free(ptr) would leak memory:
> small quantities, but potentially significant.

Ours is not the only system with this behavior, so programs with such
problems have them regardless and will be fixed if anyone cares.

> If it's one of those things that is near-impossible to change because it
> would break everything, then nevermind.

It is probably a small number of programs that would break.  But without a
much more compelling motivation, we don't go making changes that we expect
to break any programs at all.